Classification prediction | MATLAB implements PSO-CNN particle swarm algorithm to optimize convolutional neural network data classification prediction

Classification prediction | MATLAB implements PSO-CNN particle swarm algorithm to optimize convolutional neural network data classification prediction

Classification effect

Basic description

1.Matlab implements PSO-CNN multi-feature classification prediction, multi-feature input model, running environment Matlab2018b and above;
2. Optimize convolutional neural network (CNN) classification prediction based on particle swarm algorithm (PSO), the optimization parameters are, learning rate, Batch processing, regularization parameters;
3. Two-class and multi-class models with multiple feature inputs and single output. The comments in the program are detailed and can be used by directly replacing the data;
the programming language is matlab, and the program can produce classification effect diagrams, iterative optimization diagrams, and confusion matrix diagrams;
4. data is a data set, input 12 features, divided into four categories; main is The main program, the rest are function files, no need to run, data and program content can be obtained in the download area.

programming

• Method 1 for obtaining the complete program and data: Private message the blogger and redeem programs of equal value;
• Complete program and data download method 2 (direct download from the resource): MATLAB implements PSO-CNN particle swarm algorithm optimization convolutional neural network data classification prediction

%%  优化算法参数设置
SearchAgents_no = 3;                  % 数量
Max_iteration = 5;                    % 最大迭代次数
dim = 3;                              % 优化参数个数

 
%% 建立模型
lgraph = [
 
 convolution2dLayer([1, 1], 32)  % 卷积核大小 3*1 生成32张特征图
 batchNormalizationLayer         % 批归一化层
 reluLayer                       % Relu激活层

 dropoutLayer(0.2)               % Dropout层
 fullyConnectedLayer(num_class, "Name", "fc")                     % 全连接层
 softmaxLayer("Name", "softmax")                                  % softmax激活层
 classificationLayer("Name", "classification")];                  % 分类层




%% 参数设置
options = trainingOptions('adam', ...     % Adam 梯度下降算法
    'MaxEpochs', 10,...                 % 最大训练次数 
    'MiniBatchSize',best_hd, ...
    'InitialLearnRate', best_lr,...          % 初始学习率为0.001
    'L2Regularization', best_l2,...         % L2正则化参数
    'LearnRateSchedule', 'piecewise',...  % 学习率下降
    'LearnRateDropFactor', 0.1,...        % 学习率下降因子 0.1
    'LearnRateDropPeriod', 400,...        % 经过800次训练后 学习率
%% 训练
net = trainNetwork(p_train, t_train, lgraph, options);

%% 预测
t_sim1 = predict(net, p_train); 
t_sim2 = predict(net, p_test ); 

References

[1] https://blog.csdn.net/kjm13182345320/article/details/129036772?spm=1001.2014.3001.5502
[2] https://blog.csdn.net/kjm13182345320/article/details/128690229